functions to convolve simple distributions

[glennmoy]

These are simple functions to convolve distributions where a closed form exists.

Edit: taken essentially from this list: https://en.wikipedia.org/wiki/List_of_convolutions_of_probability_distributions

Please advise if this is the correct place to put them or if somewhere else would be more suitable.

[mschauer]

Nice start, next step to think of are tests.

[codecov-io]

Codecov Report

Merging #919 into master will increase coverage by 3.09%.
The diff coverage is 100%.

@@            Coverage Diff             @@
##           master     #919      +/-   ##
==========================================
+ Coverage   73.94%   77.03%   +3.09%     
==========================================
  Files         107      110       +3     
  Lines        5269     5235      -34     
==========================================
+ Hits         3896     4033     +137     
+ Misses       1373     1202     -171

Impacted Files	Coverage Δ
src/Distributions.jl	100% <ø> (ø)	⬆️
src/convolution.jl	100% <100%> (ø)
src/utils.jl	74.5% <0%> (-1.97%)	⬇️
src/multivariate/dirichletmultinomial.jl	98.55% <0%> (ø)	⬆️
src/matrix/matrixnormal.jl	97.95% <0%> (ø)
src/matrix/matrixtdist.jl	98% <0%> (ø)
src/multivariate/mvlognormal.jl	96.82% <0%> (+0.05%)	⬆️
src/univariates.jl	64.92% <0%> (+0.48%)	⬆️
src/univariate/discrete/binomial.jl	73.4% <0%> (+0.77%)	⬆️
... and 39 more

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 432545a...af97ea1. Read the comment docs.

[nickrobinson251]

thanks for adding these!

[nickrobinson251]

BTW \oplus ⊕ has been suggested as an alias for the sum of idendepent ranom Variables which is (the generalisation) of what convolve does. (Presuming we're okay with conflating RVs with their distributions).

Should we add that here?

maybe @mschauer has thoughts?
#307 (comment)

[matbesancon]

BTW \oplus ⊕ has been suggested as an alias for the sum of indedepent random Variables which is (the generalisation) of what convolve does. (Presuming we're okay with conflating RVs with their distributions).

I think convolve leads to less ambiguity, but would be ok with \oplus as alias

[matbesancon]

In any case this could go in another PR

[matbesancon]

alright, I think we can merge this as soon as travis passes, check if you want some last modifications

[glennmoy]

alright, I think we can merge this as soon as travis passes, check if you want some last modifications

just cleaned up the comments. merge when ready!